Stanford scientists invent AI-powered immune system to prevent early labor

New research shows that two common over-the-counter medications could potentially help combat premature labor, a leading cause of death in children under five.

The study, published Wednesday in the medical journal Science Advances, shows new evidence that taking low-dose aspirin combined with a common heartburn medication can reduce a pregnant woman’s risk of going into early labor. Scientists based at Stanford University conducted the study using the “digital twin” immune system through artificial intelligence models to simulate how certain medications might affect pregnancy outcomes.

In the United States, about 1 in 10 pregnant women will have preterm labor, defined as delivery before 37 weeks of gestation, and many have no warning signs before delivery. The problem is even more pronounced globally where, depending on the country, up to 16% of infants are born prematurely.

Currently, there are few options to stop early labor once it begins, but one potential cause could lie in the immune system. During pregnancy, the pregnant person’s immune system changes dramatically. First, it will weaken to allow the fetus to survive, and toward the end of pregnancy, “pro-inflammatory” markers will naturally increase in preparation for labor and delivery.

Experts say it’s these late-stage markers that are linked to an increased risk of preterm labor if they develop too early.

Dr. Brice Gaudillière, professor of anesthesiology, perioperative medicine and pain medicine at Stanford, is the study’s lead author and said his lab has been studying the immunology of pregnancy for more than 10 years.

“Over the past decade, no new drugs or pharmacological interventions have been proposed for the treatment of preterm birth,” he told SFGATE, emphasizing that clinical trials for complex conditions like preterm birth are difficult. “It’s long, it’s difficult, it’s expensive.”

In this study, Gaudillière and his team used AI models trained on data from a quarter of a billion immune cells to create what is essentially an “atlas” of the immune system. They also enrolled 11 pregnant and eight non-pregnant women in the study and took blood samples from them over a period of time. From these samples, they exposed real immune cells to different drugs. They then took this data and plugged it into the AI ​​model to see how a specific pregnant person’s immune system might respond to different medications.

“We’re creating a virtual representation of the mother’s immune system so we can first test how different drugs are going to change the immune system without giving the drug to the patient and… on top of that, simulate how the drug is going to change the outcome here,” he said.

The team found evidence that a combination of low-dose aspirin and the heartburn medication lansoprazole, known by the brand name Prevacid, showed clear signs of combating immune system dysfunction linked to premature labor.

“In the digital twin system, the combination of these two drugs had a significant effect on labor delay, 15 times greater than aspirin alone,” Gaudillière said in a statement.

Although the research is in its early stages, Gaudillière said this AI model could be beneficial for use in future studies by opening a new frontier in personalized patient care.

A drug that is not suitable for all patients is “probably the reason why drugs have failed so far in preterm birth,” he said, calling preterm birth a “complex multifactorial process.”

The team has already received approval from the U.S. Food and Drug Administration to begin a Phase 2 trial in human patients in the coming months. The team plans to use the AI-modeled data to determine who might benefit most from the drug combination, then administer the drugs to patients to see if they are less likely to have preterm labor.

Dr. Avish Arora, an obstetrician-gynecologist and maternal-fetal medicine specialist at UCLA who was not involved in the study, called the research “fascinating.”

The researchers’ approach to studying premature birth and preterm labor through the use of the digital immune system was “really novel”, he said. In many cases, doctors are unable to isolate a specific reason why a pregnant woman goes into early labor, although there are known risk factors, including a short cervix.

Arora cautioned that more research is needed to prove that these drugs will actually reduce the risk of premature birth.

“It’s definitely not causation, but it’s certainly something that’s very interesting,” he said.